Do longer lines take longer? Reconsidering the cognitive reflections of spatial duration metaphors: Evidence from Dutch

(1)

Do longer lines take longer?

Reconsidering the cognitive reflections of spatial duration metaphors:

Evidence from Dutch

Marion Baksteen (1160001) Research MA thesis Linguistics

Leiden University Faculty of Humanities Supervisors: Dr. E.I. Crevels

Dr. S. Gryllia Second reader: Dr. N.Q. Emlen

(2)

(3)

Acknowledgements

“Plans fail for lack of counsel, but with many advisers they succeed.” (Proverbs 25:22, NIV)

Written on a pink sticky note, stuck for months to the edge of my monitor, this quote finally prevails. And I am very grateful to all those who supported and counselled me during the process of writing this thesis. Specifically, I want to express my gratitude towards my supervisors Mily Crevels and Stella Gryllia. I want to thank Mily for all her insightful comments, the private classes she taught me on academic writing, and not in the least for her patience when I was still looking for the right directions and came up with different plans almost every time we met. I want to thank Stella for guiding me through the maze of what is being called statistics, I learned so much from it! Furthermore, I want to thank my father for his enthusiastic engagement in extensive discussions on my thesis, and for the invaluable effort he put into programming the experiment. I am very grateful to everyone who voluntarily participated in the experiment. I want to thank my family for encouraging me time and time again during this project. And my husband, whose support cannot be measured or described.

(4)

(5)

Abstract

The conceptual link between space and time is accounted for by two different theories: Conceptual Metaphor Theory (CMT, Lakoff and Johnson 1980) and A Theory of Magnitude (ATOM, Walsh 2003). Within a linguistic framework, CMT provides evidence for an asymmetric conceptual link between space and time, opposed to the symmetric link predicted by ATOM. Casasanto and Boroditsky (2008) and Casasanto (2010) presented evidence in favour of CMT from non-linguistic psychophysical tasks. Longer lines appeared to positively affect participants’ estimation of duration, analogous to metaphors for duration using spatial words such as long and short, but duration did not influence the perception of space (Casasanto and Boroditsky 2008). A subsequent study revealed language specific differences in effects of different stimulus types, parallel to the typical duration metaphors found in these languages (Casasanto 2010). The present investigation of Dutch shows that the relation between duration metaphors and the perception of space and duration is less straightforward than what might be expected on the basis of the accounts of Casasanto and Boroditsky (2008) and Casasanto (2010). The results of an experiment with speakers of Dutch reveal a symmetric link between space and duration in the case of space presented in the form of one-dimensional length, but an asymmetric link is reported in case of more-dimensional size. Overall, this provides evidence for ATOM rather than CMT.

(6)

List of abbreviations

1 first person 2 second person 3 third person ADJZ adjectivizer AUX auxiliary

C common (not neuter)

CONC concessive

DEF definite determiner

DEM demonstrative

DIM diminuative

DIST distal

F feminine

FUT future

INDEF indefinite determiner

INF infinitive M masculine N neuter NEG negative NMLZ nominalizer PTCL particle PL plural

POSS possessive pronoun

PP perfect participle PRO pronoun PRS present QUANT quantifier PST past REC reciprocal

REFL reflexive pronoun

(9)

List of figures and tables

Figure 1: Top 10 most frequent duration metaphors in SoNaR………..19

Figure 2: Screenshots of stimulus presentations………42

Figure 3: Mean estimations of duration and space values.………47

Figure 4: Mean estimations as percentage of stimulus value………..48

Table 1: Top 10 most frequent duration metaphors in SoNaR………...18

Table 2: Ratio of types of ‘lang + noun’……….…25

Table 3: Ratio of tokens of ‘lang + noun’………...…25

Table 4: Frequency distribution of 200 ‘lang + noun’ types………...…26

Table 5: Ratio of types of ‘kort + noun’……….26

Table 6: Ratio of tokens of ‘kort + noun’………...…26

Table 7: Frequency distribution of 200 ‘kort + noun’ types………...…27

(10)

(11)

1. Introduction

1.1. Time in language

Time is a feature of great interest in linguistics. It is intriguing that an intangible phenomenon such as time is represented so strongly in the grammars of human languages. Over centuries, linguists have described TAM systems for many languages, systems that reflect temporal experiences of the speakers of these languages. Another interest regarding time and language is found in the study of how time as a concept is represented in linguistic utterances. Well-known metaphors, such as ‘time flies when you’re having fun’, reflect our ideas about time, and possibly shape them at the same time. Moreover, not only do we talk about the concept of time as such in metaphors, but a whole range of constructions we use to discuss temporal arrangements make use of metaphors as well. Take for example the way we tend to talk about future events as ‘coming events’, or about the past as ‘laying behind us’. Though a maybe less well-studied subject than the grammatical encoding of time, there exists a vast amount of literature on this matter, not only from a linguistic perspective, but for example also from psychological and philosophical perspectives. Relatively new in this area of study is the interest in the link between time and space. Though it has long been observed that many languages discuss time in terms of space, this notion has only relatively recently begun to be investigated in a structured way.

The current thesis is concerned with this link between space and time, in languages, but also with respect to the cognitive reality of such a link. Specifically, attention will be paid to so-called ‘duration metaphors’ in Dutch, expressions that find their source in the semantic domain of space, but that describe duration. The thesis also reports on an experiment which tests possible effects of such metaphors on the perception of space and duration. However, before turning to that, it is important to be aware of certain developments in this field of investigation.

1.2. Time and Space

One of the pioneering studies regarding the link between time and space is found in the work of Clark (1973). In this paper, Clark argues that English time expressions are based on a spatial metaphor. He proposes two spatial metaphors of time: moving ego and moving time (Clark 1973:50). In explaining these metaphors, Clark compares time to a highway of discrete events in a successive order. The different perspectives one can take with respect to this highway result in the two different metaphors. According to one perspective, speakers (the ego) are moving along this highway, so that future time is ahead of the ego and past time behind the ego, resulting in the moving ego metaphor. An example Clark gives of this metaphor is:

(1) We are just coming into troubled times. (Clark 1973:51)

The other perspective reflects the highway of time as moving past the ego, from front to back, resulting in the moving time metaphor, as in:

(2) Friday arrived before we knew it. (Clark 1973:50)

(12)

In the years following this paper, the notions of moving ego and moving time were taken up by many scholars in the description of spatial metaphors of time. An important impulse to the development of more specific theories on such metaphors was given by the work of Lakoff and Johnson (1980, 1999) and Lakoff (1993), who initiated the framework of Conceptual Metaphor Theory (CMT).

In CMT, the metaphorical link between a target domain, for example ‘time’, and the source domain from which it receives its terminology, for example ‘space’, is thought to reflect a real link in cognitive sense. Metaphors found in daily use are argued to reflect the way the concepts discussed are conceptualised. In the tradition of CMT, metaphors are usually represented in (small) caps, such as: TIME PASSING IS MOTION (Lakoff 1993:14). This metaphor is said to account for a wide range of spatial expressions used for time. Two special cases of this metaphor are proposed: TIME PASSING IS MOTION OF AN OBJECT and TIME PASSING IS MOTION OVER A LANDSCAPE. These metaphors cover respectively the moving time and moving ego

metaphors of Clark, but are broader. To illustrate, see the following examples: (3) The time will come when…

(4) Let’s put all that behind us. (Lakoff 1993:14)

Both (3) and (4) are instances of TIME PASSING IS MOTION OF AN OBJECT. Of these two examples

only (3) would probably also be classified as moving time. On the other hand, (4) cannot without difficulty be regarded as moving time. In (4) there is an aspired movement of some temporal event/experience, described as ‘all that’, towards the backside of the ego. However, this movement is not the natural flow of time from front to back as perceived by an ego. Instead, this aspired movement is induced by the ego itself, as is understood from ‘let’s put’. Hence it cannot really be stated that example (4) is an instance of a moving time metaphor. Something similar can be said for TIME PASSING IS MOTION OVER A LANDSCAPE compared to moving ego: (5) We’re coming up on Christmas.

(6) His stay in Russia extended over many years. (Lakoff 1993:14)

While (5) and (6) are both examples of TIME PASSING IS MOTION OVER A LANDSCAPE, only (5)

can be regarded as moving ego without difficulties. In (6) we cannot really speak of a moving

ego as it literally states that the ego is staying in the same place. Thus, while the metaphors for

time as movement proposed by Clark (1973) and Lakoff (1993) partly overlap, they cannot be used interchangeably.

In this way, different linguists proposed various adapted, complementing or new metaphors based on their own data for different languages, and the body of literature on this matter grew (see for example Alverson 1994, Yu 1996, Dahl 2005). Specifically with respect to the frames of reference proposed by Levinson (2003) for spatial orientation in languages, studies appeared applying this to temporal metaphors (Bender et al. 2005, Núñes and Sweetser 2006, Moore 2006, 2011, 2014, Bender et al. 2010, Boroditsky and Gaby 2010, Shinohara and Pardeshi 2011, Brown 2012, Fedden and Boroditsky 2012, Gaby 2012, Núñes et al. 2012). Based on such studies, several linguists have proposed a taxonomy of temporal frames of references (see for example Zinken 2010, Tenbrink 2011, Evans 2013a, 2013b). For an

(13)

extensive review of the state of the art in this field of investigation, I refer to the comprehensive work of Bender and Beller (2014).

Not surprisingly, alongside the investigation of such conceptual metaphors and temporal frames of reference in different languages, interest rose in the question to what extent we can speak of a cognitive reality of a spatial representation of time. Many of the mentioned studies already incorporate part of that question in the discussion of their findings. Other studies were designed specifically to give insight in this question. A popular task used in such studies involves the ambiguous expression ‘Next week Wednesday’s meeting has been moved forward two days’, in which people have to respond to the question on which day the rescheduled meeting will take place. The two possible interpretations of this expression reflect either a

moving time or a moving ego perspective. From a moving time perspective the correct answer

would be Monday, as the forward movement of time implies its coming closer to the ego. From a moving ego perspective the correct answer would be Friday, as the movement of ego means that moving forward implies moving further into the future.

As said, this task was implemented in different studies. Boroditsky and Ramscar (2002) for example found that the answer people would give could be manipulated by priming them with a picture task involving either movement of an object towards the ego or movement of the ego towards an object. Likewise, the real experience of motion also seemed to influence the answer people would give. A questionnaire amongst people in an airport revealed that people who had just flown in were significantly more likely to answer from a moving ego perspective than people who were about to fly. Also, people who were about to fly were more likely to answer from a moving ego perspective than people who were at the airport to pick someone up (Boroditsky and Ramscar 2002:186).

A side note to this finding is that within the ‘picking up people at the airport’ group there was no significant difference between the number of people who answered from a moving

time perspective and the ones that answered from a moving ego perspective. So, although you

might expect this group of people to be more likely to conceive time as coming towards them instead of moving through time themselves, as they are waiting for people coming towards them, there is no evidence for this in the data. Actually, it might be the case that for English, in a neutral condition a moving time and moving ego perspective are equally likely to occur, and that experiences of (anticipated) motion can only facilitate the moving ego perspective. On the other hand, the lack of motion as experienced by the ego that is implied in the moving time perspective, does not seem to be a sufficient basis for influence of an ego’s experience on the choice of perspective. The study of Boroditsky and Ramscar (2002) is but one out of many studies on the cognitive reality of a spatial representation of time as found in language. Others include for example Matlock et al. (2005), Fuhrman et al. (2011), Bender et al. (2012), Lai and Boroditsky (2012), and de la Fuente et al. (2014). However, not every experiential link between space and time is explained by linguistic metaphors, see for example Bergen and Lau (2012), and Sousa (2012) on the influence of writing direction on the way people map space onto time. So far, the mentioned studies all discuss the linguistic relation between time and space from the perspective of temporal frames of reference. In other words, how different events relate to one another in temporal respect, or how particular events relate to the temporal ground of the speaker (the ego). But the semantic domain of time covers a much greater range of topics.

(14)

Haspelmath (1997) summarizes claims on the link between temporal and spatial expressions found in previous linguistic studies as follows:

(i) Temporal expressions are identical with spatial expressions. (Wierzbicka 1973, Clark 1973, Jackendoff 1983)

(ii) Temporal expressions are based on spatial expressions.

(Meyer-Lübke 1899, Gamillscheg 1957, Lyons 1977, Langacker 1987, Wunderlich 1985)

(iii) Speakers conceive of time in terms of space. (Gamillscheg 1957, Langacker 1987)

(Haspelmath 1997:18)

There is a hierarchy in these claims, with (i) having the least and (iii) the most consequences. In this list, Haspelmath does not refer to CMT, but the work of Lakoff and Johnson (1980, 1999) best fits the claim in (ii) and to a certain extent also the claim in (iii). CMT does hypothesize that speakers conceive of target domains in terms of the source domain from which they get their metaphors, but the CMT framework does not provide directions to test such hypotheses. A theory that is often mentioned as opposing CMT is A Theory of Magnitude (ATOM), as proposed by Walsh (2003). According to ATOM, time, space, and quantity (also referred to as number) are part of a domain-general magnitude system. Where CMT stems from linguistics, ATOM is based on brain studies. For an extensive discussion of the differences between CMT and ATOM, and a review of the evidence for either of the two theories, I refer to Winter et al. (2015).

The main difference between CMT and ATOM is the way they view the (a)symmetry between the domains of space, time, and quantity. This (a)symmetry is found on domain as well as directional level. Domain (a)symmetry refers to the extent to which different domains are linked. Directional (a)symmetry refers to the extent to which two linked domains influence each other. Overall, ATOM advocates domain symmetry as well as directional symmetry. In other words, according to ATOM quantity and time are as likely to be linked as space and time (domain symmetry), and space might influence time to the same extent as time influences space (directional symmetry). CMT on the other hand supports an asymmetric view. According to CMT, there are source domains and target domains, target domains do not influence source domains to the same extent source domains influence target domains1 (directional asymmetry). A link between different target domains is not necessary for CMT (domain asymmetry), and does not exist the case of number and time. According to CMT, based on linguistic evidence, the only link time and number might display, goes via the domain of space (see Winter et al. 2015).

Winter et al. (2015) discuss evidence for domain and directional (a)symmetry from both perspectives, and conclude that, unlike the way they are often presented, CMT and ATOM are not completely incompatible theories. Precisely because evidence for both theories usually

1_{Asymmetry is distinguished from unidirectionality; there are examples of time expressions metaphorically used} to indicate space (e.g. ‘I am five minutes away from the library’), but they are claimed to be less frequent than spatial expressions used to indicate time.

(15)

comes from different disciplines and focuses on different aspects of the links between the domains, the theories might complement each other. In this respect, they emphasize the focus of ATOM on low level perception, and the focus of CMT on the understanding and reasoning about complex concepts (Winter et al. 2015:220).

1.2.1. Duration and Space

So far I have discussed two theoretical frameworks, CMT and ATOM, in the light of the link between time and space. Time is a very broad concept, and thus far, the focus has merely been on the structuring of events in time, and the way languages rely on the domain of space in describing such events. As mentioned, there exists a vast amount of linguistic literature on this topic, reviewed by Bender and Beller (2014). The relation between duration and space, on the other hand, has been studied from different disciplines, but has received far less attention from linguists than the structuring of events in time. From the perspective of ATOM, it is very likely that a link exists between space and duration, since duration is by definition that aspect of time that has to do with magnitude. Also from the perspective of CMT, it is likely that space functions as a source domain for duration metaphors. With respect to that, the study of Galton (2011) on shared attributes of space and time gives useful insights.

Galton (2011) argues time has four key attributes: extension, linearity, directionality and transience. Of these four attributes, three are shared with the domain of space: extension,

linearity, and directionality. Based on these shared attributes, space can function as a source

domain for time metaphors. With respect to extension, linearity, and duration, Galton writes: “The extendedness of time seems to consist in the fact that its nature is such that

what is in time […] can be separated, so that even things that are otherwise identical, such as the same action by the same subject in the same place, can be numerically distinct solely by virtue of occupying distinct times. Duration might then be regarded as a measure of the extent of their separation, but it does not seem to be possible to define this without reference to what comes between the two times – a concept that relates to linearity rather than bare extension.”

(Galton 2011:697)

In other words, there is an aspect of time that is best described as extension, which makes it possible to distinguish between different moments. The concept of duration, which in itself might be regarded a sub-concept of the broader concept of time, is linked to this attribute of time. Yet according to Galton, duration requires another attribute of time, namely linearity: the fact that time is one-dimensional2_{. This conclusion appears to be based on an English}

perspective on duration metaphors; elsewhere in his paper, Galton mentions the use of long or

short to refer to duration as an example of spatial metaphors for time exploiting the attribute of

linearity (Galton 2011:700). However, this might not be a universal way to describe duration. As opposed to duration, the structuring of events in time relies more on the attributes of linearity and directionality instead of extension.

2_{With respect to the discussion on ATOM and CMT it is interesting that Galton, in his attempt to define the} attributes of time without using spatial terminology, resorts to terminology from the domain of quantity/number. For example, he defines linearity as: “relative to a given moment of time, it requires only one number to specify the position of another moment” (Galton 2011:697).

(16)

I have not found any highly detailed accounts on metaphors for duration from a CMT perspective. Though various studies discuss the effect of space on duration estimation (see for example Xuan et al. 2007, Casasanto and Boroditsky 2008, Bottini and Casasanto 2010, Casasanto et al. 2010, Srinivasan and Carey 2010, Cai and Connell 2015), an extensive description of the metaphors on which the experimental set-up is based, is often no core goal of the study. According to the positions they take on the link between language and the conceptualisation or perception of space and duration, the mentioned studies can be classified in three different groups:

(i) No consideration of linguistic metaphors at all. (Xuan et al. 2007)

(ii) Linguistic metaphors are brought up as conceptual metaphors influencing participants’ perception of space and duration.

(Casasanto and Boroditsky 2008, Casasanto 2010, Bottini and Casasanto 2010, Casasanto et al. 2010, Srinivasan and Carey 2010)

(iii) Linguistic metaphors are considered as irrelevant for participants’ perception of space and duration.

(Cai and Connell 2015)

Xuan et al. (2007) concluded on the basis of different stimuli, for example pictures of few and pictures of more dots, that the size of a stimulus affects the estimation of duration. However, since no link to linguistic metaphors is made, I will refrain from further discussion of this paper for the current moment (but see footnote 23, chapter 4). The study of Xuan et al. (2007) is just one example to illustrate this group of non-linguistic studies; many more have been published, often with reference to ATOM (see Winter et al. 2015 for references).

The studies in group (ii) are based on CMT, their aim is to test claims made by CMT that we perceive abstract concepts in the way linguistic metaphors describe them. Casasanto and Boroditsky (2008) describe several experiments including visualised spatial displacement, they report a positive correlation between displacement and duration estimation, and link this to English referring to durations as either long or short. Casasanto (2010) builds forth on these experiments by designing a different kind of stimulus to match languages that use other spatial terminology to refer to duration. On the basis of experiments carried out among participants speaking different languages, Casasanto (2010) concludes that the language a participant speaks determines which type of stimulus affects most the perception of duration. In another study, Casasanto et al. (2010) found similar asymmetric relations between time and space in children carrying out duration and space estimation tasks. This led them to the conclusion that CMT gives a better account of the relation between time and space than ATOM (Casasanto et al. 2010:403)

Bottini and Casasanto (2010) specifically contrast ATOM and CTM, favouring the latter on the basis of two experiments with speakers of Dutch. In the first experiment, participants had to estimate the duration of stimulus words of which the semantics referred to different spatial lengths (e.g. pencil and footpath); results showed an effect of implicit spatial length on duration estimation. In the second experiment, people had to estimate the exact spatial length of stimulus words referring to events with different durations (e.g. blink and season); results

(17)

showed no effect of implicit duration on the estimation of spatial length. This led Bottini and Casasanto (2010) to the conclusion that the link between space and time is asymmetric in direction; space influences time perception, but time does not influence space perception, congruent with the pattern found in metaphors from space as source domain to time as target domain.

Based on experiments with congruent and incongruent space/time stimuli (short stimuli with long durations, long stimuli with short durations), Srinivasan and Carey (2010) also report that participants link spatial length and duration. The experimental design did not allow for conclusions on the direction of this link. However, since they found the same effect in 9 month old infants, they concluded it is not possible to explain the way humans link space to time based on metaphors found in language. Rather, they support the view that language might influence the link that is already perceived by humans before any language is acquired.

Contrary to these studies from group (ii), Cai and Connell (2015) favour ATOM over CMT. Based on experiments with stimuli of which the spatial features were perceived through different senses, they put forward a different view on the relation between space and time. Rather than assuming a reported asymmetric relation in which space influences time accounts for the complete domains, they focus on differences in acuity of perceptual modality. In that way, they found that duration affects space perception if space is perceived through touch (low perceptual acuity). On the other hand, space affects the perception of duration if space is perceived through vision (high perceptual acuity). This effect might explain all the effects reported in the studies from group (ii) discussed above, since in all those cases, space was perceived visually.

The study of Cai and Conell (2015) shows that what might seem clear evidence of asymmetry in direction, and thus a support for CMT rather than ATOM, might actually be explained in a different way. This is a reminder that support for a hypothesis is not necessarily evidence that the hypothesis is true (correlation does not equal causation). Yet, conclusions on implications of found effects are rather far reaching for some of the discussed studies. In particular Casasanto links his investigations of the relationship between space and duration to the debate on linguistic relativity (Casasanto 2008:70-75, Casasanto 2010, Casasanto 2016:160-162).

The discussion on linguistic relativity, often referred to as the Sapir-Whorf hypothesis, is concerned with the relation between language and thought. The extreme version of this hypothesis is that language and thought are inseparable, in which language determines and limits speakers’ thoughts (based on von Humboldt 1988 [1836], Sapir 1924, Whorf 1956). This view since long is rejected by most linguists, but moderate versions of the theory are gaining ground. An early moderation of the idea of linguistic relativity, is the nuanced version that language, specifically with respect to grammar, does not so much determine what speakers can pay attention to, but rather what they must pay attention to (see e.g. Boas 1938:132-133, Jakobson 1959:236, Slobin 1996). On the other hand, studies on gender, colour terms, and space reveal that language to a certain extent does influence speakers’ perception of the world.3 And

3_{For an extensive discussion on the development of theories concerning linguistic relativity and recent} anthropological linguistic research into this matter, I refer to the popular, accessibly written, account of Deutcher (2010).

(18)

related to space, time is now also being presented as perceptually influenced by language. This latter claim is the focus of the present thesis.

1.3. Research question, method and terminology

Often, the studies of Casasanto and Boroditsky (2008) and Casasanto (2010) are the only studies referred to as evidence for an asymmetric link between space and time, based on conceptual metaphors. Yet a critical review of these studies is lacking. The aim of the current thesis is twofold. On the one hand it strives to provide a critical review of Casasanto and Boroditsky (2008) and Casasanto (2010), with a focus on methodology. On the other hand, a case study on Dutch, consisting of a corpus investigation and a psyhophysical experiment, is carried out to provide new evidence on the relation between linguistic metaphors and the conceptualisation of space and time. The main research question is:

Does evidence from Dutch confirm the findings of Casasanto and Boroditsky (2008) and Casasanto (2010) that the link between time and space is asymmetrical, congruent with patterns in linguistic metaphors?

Note that I am hesitant to use the term conceptual metaphor. Rather, I prefer to speak of

linguistic metaphors, until conclusive evidence is provided that these metaphors indeed reflect

deeper cognitive conceptualisations.

One of the aims of the section 1.2 was to illustrate the diversity in studies on space and time in linguistics and beyond. Not surprisingly, the terminology for specific core concepts in this field of investigation is almost equally diverse. For the sake of clarity, in this section, I list the working definitions I use for several concepts that are referred to throughout the thesis. Whenever relevant, other concepts are defined when introduced in the different chapters.

Time metaphor

A linguistic utterance about (a part of) the semantic domain of time including terminology or constructions characteristic for another semantic domain. Spatial time metaphor

Time metaphor including terminology/constructions from the semantic domain of space.

Temporal frame of reference

Spatial time metaphor that places (a) temporal event(s) in a certain position with respect to either (an)other temporal event(s) or (an)other entity(/entities). Duration metaphor

Spatial time metaphor about the time span of a temporal event, or about duration in general.

Duration metaphors also occur as non-spatial time metaphors, such as in a good hour, indicating an hour that is felt to take long. Though this expression is metaphoric, it does not include spatial terminology. Since these metaphors are not considered within the scope of this thesis, duration

metaphor always refers to spatial duration metaphors unless indicated otherwise.

The remainder of this thesis is structured as follows. Chapter 2 introduces the papers by Casasanto and Boroditsky (2008), and Casasanto (2010) that led to the present investigation of

(19)

Dutch duration metaphors. Though the main points of these papers are summarized in this chapter, the focus will be on the linguistic analysis that formed the basis for the different experiments of these studies. In the following chapter, chapter 3, suggestions from chapter 2, on improvements for linguistic investigation of duration metaphors, are applied in a corpus-based study of Dutch duration metaphors. On the basis of this analysis, chapter 4 returns to Casasanto and Boroditsky (2008) and Casasanto (2010) in a discussion of their implementation of linguistic findings in psychophysical tasks. This chapter ends with a proposal for implementation of the findings of the Dutch linguistic investigation of chapter 3 in an experimental setting to test the relationship of these linguistic features with the perception of time and space. Finally, chapter 5 reports on this experiment, which was carried out among 20 native speakers of Dutch. A general conclusion is found in chapter 6.

(20)

2. Theory: linguistic investigation

2.1. The studies

The present investigation of Dutch duration metaphors is in reaction to the studies of Casasanto and Boroditsky (2008) and Casasanto (2010). The starting point of Casasanto and Boroditsky (2008) is that in languages, the relation between time and space is asymmetric. They argue that English exhibits much more expressions of time from which the terminology is borrowed from the domain of space, than expressions of space that borrow terminology from time. An example would be “a long vacation and a short concert” (Casasanto and Boroditsky 2008:580). Though not explicitly stated, from these examples it is clear that Casasanto and Boroditsky are concerned with duration metaphors rather than temporal frames of reference. Based on this premise, they hypothesize that, if the relation between time and space goes beyond language use and is also reflected in our non-linguistic thinking of time and space, we would expect an asymmetric relationship there as well. To test this, they developed the growing line task, a task in which participants view a line that is horizontally growing on a screen. At a certain point the line disappears, after which participants either have to estimate the maximal length of the line they just saw, or the timespan that the line took to grow.

Six different variants of this experiment were carried out4: 1. Growing lines, as described above.

2. Growing lines, selective attention. People were told before every stimulus whether they had to answer the space- or the time-question afterwards.

3. Growing lines, temporal frame of reference. The time frame in which the stimuli were shown, were proportional to the length of the line and its growing speed.5

4. Growing lines, concurrent tone. The stimulus time was not only made available through the growing time of the line, but also through a constant tone that accompanied the growing time of the line.

5. Moving dot. Similar to growing lines, but instead of a growing line, a dot was shown that moved horizontally along the screen. For the space-question, participants had to indicate the starting and finish point of the moving dot.

6. Stationary line. Instead of a growing line, a stationary line was displayed on the screen for a certain time.

All six experiments roughly showed the same results. In all six, the length of the line (or the length of the path, in case of the moving dot) influenced the estimation of duration. On the other hand, the actual duration did not influence the estimation of length.

Although these are very interesting findings, some critical comments might be in order. First, there is the issue of growing speed. By using a growing line, or a moving dot, more information is provided than merely duration and length, namely the growing speed of the line.

4_{Based on Casasanto and Boroditsky (2008); an earlier report of these experiments is found in Casasanto and} Boroditsky (2003), based on Casasanto’s doctoral dissertation (2005).

5_{It is not completely clear from their explanation what Casasanto and Boroditsky mean by this. Apparently, before} and after the presentation of each line, a period of delay was inserted. In other words, the total duration of a stimulus became the time the line would have taken to grow from the left edge towards its actual starting point plus the time it would have taken to grow from its final point towards the right end of the screen, plus the time the line actually was growing on the screen.

(21)

The analysis of the data did not include a test of the effect of growing speed on either estimation of time or estimation of space, this point will be discussed in more detail in chapter 5. However, it is striking that the effect of length on duration estimation in experiment 6, the only experiment that did not include growing speed, though still significant, is less significant than in the other five experiments (cf. Casasanto & Boroditsky 2008:581-587). A maybe even more striking detail is the number of participants that did the different experiments. This ranges from 9 to 19 people per experiment. It is interesting that the experiment in which 19 people participated, is experiment 6, the one with the less significant results.

In another study, Casasanto (2010) builds forth on the findings of Casasanto & Boroditsky (2008), by linking performances to linguistic backgrounds. Comparing temporal metaphors in different languages, he distinguishes two different types of metaphors for duration: Distance Metaphors, such as a long time, and Amount Metaphors, such as much time. For these two expressions, Casasanto elicited equivalents from native speakers of French, Spanish and Greek. Subsequently, these expressions were inserted as a search term in Google, and the number of hits was noted. It appeared that for French and English, the distance metaphor was much more frequent (more than 70% of all instances) than the amount metaphor. On the other hand, for Greek and Spanish it was the other way around and the amount metaphor was much more frequent (more than 80% of all instances) than the distance metaphor (Casasanto 2010:467). Based on these frequencies, Casasanto classified the different languages as either distance metaphor preference or amount metaphor preference.

When reconsidering the growing line task, it is possible that a linguistic preference for either distance or amount metaphors could influence the result, as the growing line task is clearly about distance rather than amount. To check whether evidence could be found that speakers of a distance metaphor language perceive time differently than speakers of an amount metaphor language, another task was designed. As an amount metaphor counterpart of the growing line task, Casasanto developed the so called ‘filling tank task’. As the name says, in this task people viewed a schematically drawn tank gradually filling with water. After each stimulus, participants either had to indicate how full the tank had become or how long they had seen the tank being filled. It was predicted that speakers of an amount metaphor language would show a stronger effect of space on duration estimation for the filling tank task than the growing line task. Speakers of distance metaphor languages, on the other hand, are expected to be more influenced by space in the estimation of duration, when participating in the growing lines task rather than the filling tank task. These hypotheses were confirmed when the results of the two experiments were compared for Greek (amount metaphor) speaking and English (distance metaphor) speaking participants (Casasanto 2010:469-471).6

The findings of Casasanto (2010) are even more far reaching than those of Casasanto and Boroditsky (2008). Both studies conclude that there is a directionally asymmetric link between space and time in our non-linguistic conceptualization of time as well as in the languages we speak. But besides that, Casasanto (2010) concludes that the way a particular language presents the link between space and time, influences the way time is perceived by speakers of that language. Or, in his own words:

6_{A preliminary report on this experiment and the accompanying linguistic investigation was published by} Casasanto et al. (2004), and is found in Casasanto’s doctoral dissertation (2005) as well.

(22)

“Results are incompatible with the Shallow View of language-thought relations, and provide some of the first evidence for the view that language has Deep influences on nonlinguistic mental representation […]” 7

(Casasanto 2010:474-475)

With this conclusion, Casasanto (2010) is skating on thin ice, entering the field of linguistic relativity, a phenomenon which mere existence is highly debated. If his conclusions are valid, they have implications for the status of linguistic features as predictors of cognitive conceptualizations. And, indeed, the results are quite impressive at first sight. Summed up, Casasanto’s findings are that:

1. Speakers of ‘distance metaphor languages’ exhibit significantly more distance interference in the estimation of duration than amount interference.

2. Speakers of ‘amount metaphor languages’ exhibit significantly more amount interference in the estimation of duration than distance interference.

3. Training8 of English speakers in either distance or amount metaphors resulted in even stronger effects when executing the filling tank task. Participants trained in distance metaphors showed less amount interference than untrained participants. Participants trained in amount metaphors exhibited significantly more amount interference than participants trained in distance metaphors (Casasanto 2010:471-473).

It seems indeed likely to assume that these results provide evidence for a Deep View on the influence of language on thought. However, a closer examination of the premises and methodology of these studies, might provide a new perspective on the reliability of these results and conclusions. Specifically the classification of languages as either preferring amount metaphors or distance metaphors is crucial for the conclusions Casasanto draws with respect to deep influences of language on mental representations. Therefore, this specific aspect of Casasanto’s (2010) report will be reviewed in the remainder of this chapter.

2.2. Distance and amount metaphors

As discussed in the previous section, Casasanto (2010) uses the terms distance and amount

metaphor to describe two different types of metaphors languages use in the description of

durational time. These metaphors are also referred to as 1-Dimensional metaphors and

3-Dimensional metaphors, respectively. However, the line of reasoning Casasanto provides for

this dichotomy is slightly confusing. He starts the discussion of different types of metaphors for duration by providing the following examples from Greek and their English equivalents:

7_{For an explanation of what exactly is meant by Shallow and Deep View, see Casasanto (2010:460-461). In short,} the Shallow View represents the idea that linguistic structures merely influence thinking that includes language, whilst the Deep View represents the idea that language also influences non-linguistic thinking.

8_{Participants had to fill in blanks in 192 comparative sentences, of which half were about duration of events and} half about features of physical objects. Participants being trained in amount had to choose between more and less, and participants being trained in distance between longer and shorter (Casasanto 2010:472).

(23)

(7) Greek (Indo-European) a. megali nychta big night ‘long night’ b. megali schesi big relationship ‘long relationship’ c. parti pou kratise polý9 party REL last.PST much

‘long party’

d. synantisi pou diekese polý meeting REL last.PST much

‘long meeting’

(Casasanto 2010:467)10

Based on these examples, Casasanto argues that where English uses distance metaphors, Greek rather expresses duration “in terms of 3-dimensional size or amount” (Casasanto 2010:467). Indeed, I think it is a fair conclusion that (7a) and (7b) express duration using size terminology, and (7c) and (7d) using amount terminology. However, directly after this observation, both expression types are subsumed under the cover term amount metaphor without further explanation. The remainder of the paper does not mention a word on a difference between size and amount in durational metaphors and possible implications for the conceptualization of time. I believe this lack of explanation is possibly due to a superficial examination of the linguistic material. As Casasanto is trained in Brain & Cognitive Sciences, his expertise in this study is mainly reflected in the investigation of the conceptualization of time and not in the analysis of the languages of which speakers are investigated. For example, it appears that, probably because Greek polý is translated in English as ‘much’ in examples (7c) and (7d), Casasanto assumes that amount metaphors always concern ‘mass’ cases. This is clear from the fact that in the experiment he makes uses of a schematic drawing of a container being filled with water, which is a mass noun. Linguistically, time is a mass-noun, but since we deal with a very abstract concept, it is problematic to assume that conceptually time is a mass entity as well. Especially when considering the duration aspect of time, which is usually referred to in distinct countable units, such as days, minutes, hours, etc. From a linguistic point of view, a legitimate question to ask would be whether Greek indeed uses amount metaphors in this ‘mass’ sense, or that it could also be a more ‘count’ sense.

9_{Casasanto uses the obsolete transcription of Greek upsilon, transcribing it with i. However, I use the modern} standard transcription y, to avoid confusion with iota, which is also transcribed as i.

10_{In this discussion, the examples are presented as if Greek was the origin for the comparison between English} and Greek. However, the way Casasanto presents these data and the fact that all English equivalents are noun phrases, while (7c) and (7d) are relative clauses in Greek, gives the impression that English was the base language and the Greek equivalents were elicited. This is a relevant difference, as it might indicate a bias towards English. Maybe Greek exhibits even more different types of duration metaphors, but they might not be provided when speakers of Greek are asked to give the best Greek equivalent for a specific English expression.

(24)

And indeed, when looking into this question, it appears that there is a difference between English and Greek in this respect. English distinguishes between two quantifiers: much for mass nouns and many for count nouns. Greek on the other hand, uses one adjective, polýs, in combination with mass as well as count nouns. In adverbial position, the neuter singular form of this adjective, polý, is used (Holton et al. 1997:80, 92). It is thus questionable whether the filled container task is an appropriate measurement to test the influence of amount metaphors on the conceptualization of time.

Besides the problem of the questionable implementation of the amount metaphor in the filled container task, the categorization of languages as having a preference for amount metaphors or distance metaphors is problematic as well. Firstly, as discussed, Casasanto subsumes under amount metaphor both metaphors denoting size in the sense of ‘a specific entity’s measurement’ as well as amount in the sense of ‘a certain quantity of distinct entities’. In other words, no distinction is made between mass amount and countable amount in the labelling of linguistic expressions. The way in which Casasanto examined the preference for the different metaphors is maybe even more problematic. When considering the expressions in example (9) it is obvious that all denote some sense of durational time, even though none actually includes a word that explicitly means time or duration. Remarkably, after providing this evidence that durational metaphors do not necessarily include the word ‘time’, Casasanto, without further explanation, merely checks frequencies of two expressions: long time and much

time (Casasanto 2010:268). However, it does not necessarily follow that, if the equivalent of

one of these expressions is more frequent in a language than the equivalent of the other expression, all durational expressions in this language follow that pattern. Theoretically, it might very well be possible that the frequency of long time as an expression in a certain language as compared to the frequency of much time is not representative for the ratio of distance versus amount metaphors as a group.

Casasanto carried out his linguistic investigation by counting Google hits for the translation equivalents of long time and much time in four languages (English, French, Greek and Spanish). From a linguistic point of view, the internet is generally not considered the most ideal corpus for an investigation how a certain expression is used in a particular language (see also Everett 2013:125). Frequency counts carried out by the Google search engine are no reliable predictors of frequency in use. The internet contains much duplication of the same texts, so one actual instance of use will be counted several times. Besides, different text types are not balanced, nor is every type of language use represented. In short, the strong preferences Casasanto found for either distance or amount metaphors in the four languages he investigated could be questioned. An evaluation of the way in which Casasanto transferred the linguistic findings to a test condition to investigate the influence of these linguistic features on cognition also gives rise to some interesting points for discussion. But before turning to that topic, some further issues that might arise when trying to classify a language on the basis of duration metaphors are considered in a discussion of Dutch data.

(25)

3. Duration metaphors in Dutch

3.1. Language and data

Dutch (Indo-European, Low Franconian), is spoken by approximately 15,700,000 people in the Netherlands (European Commission 2012), of which it is the national language. Together with the speakers from Aruba, Belgium, the Caribbean Netherlands, Curacao, Sint Maarten and Suriname, the total population of Dutch consists of 22,040,690 speakers (Lewis et al. 2016). It has SVO word order in main clauses, with SOV word order in subordinate clauses. Its inflectional morphology is strongly suffixing, noun phrases are head final with prepositions (for a detailed typological profile see Dryer and Haspelmath 2013). Though being a very widespread language, with a long tradition of linguistic investigation, as far as I know, no systematic account of temporal metaphors exists for Dutch. In the present study, a first attempt towards such an account is made. Since this study is a reaction to the ones by Casasanto and Boroditsky (2008) and Casasanto (2010), the focus is on duration metaphors.

The data that form the basis of this description are from two reference corpora: the Stevin Nederlandstalig Referentiecorpus (SoNaR) and the Corpus Gesproken Nederlands (CGN). SoNaR is a corpus of contemporary written Dutch, containing 500 million words (Oostdijk et al. 2013). It includes written to be read as well as written to be spoken material from a wide variety of text types from different conventional and new media. The corpus was developed between 2008 and 2011, in a project that was carried out by different universities from the Netherlands and Belgium, coordinated by Radboud University. SoNaR is available online via OpenSoNaR. CGN is a smaller corpus of annotated spoken Dutch. It contains about nine million words, from which approximately a third were recorded in Belgium and two thirds in the Netherlands. The corpus is not available online, but a free license is available for scientific purposes. Since pronunciation forms no part of the current investigation, merely the annotations (Nederlandse Taalunie 2014) were used.

3.2. Methodology

In the description of Dutch duration metaphors, three different types will be distinguished: 1. Distance metaphors

2. Size metaphors 3. Amount metaphors

This distinction is based on the previous evaluation of distance and amount metaphors as discussed by Casasanto. Duration metaphors are described in this thesis in three different groups rather than as a whole, because implications for space-time relations can be easily connected to these three groups: distance metaphors might facilitate the effect of spatial length on duration estimation, size metaphors might facilitate the effect of physical largeness on duration estimation, and amount metaphors might facilitate the effect of number of distinct entities on duration estimation. The aim of this section is thus to determine to what extent evidence for such a tripartite division of duration metaphors can be found in Dutch. Distance metaphors are defined similarly to Casasanto’s distance metaphor; as a working definition, I will use:

(26)

Distance metaphor

A distance metaphor is a duration metaphor containing terminology that is also used to talk about 1-dimensional space, or length.

Where Casasanto uses amount metaphor as the ‘more-dimensional’ counterpart of distance metaphor, I distinguish between two different types, based on the discussion in 3.4.2; I will use the following working definitions:

Size metaphor

A size metaphor is a duration metaphor containing terminology that is also used to talk about physical dimensions of specific entities.

Amount metaphor

A number metaphor is a duration metaphor containing terminology that is also used to talk about physical amounts of distinct entities, including quantifiers.

In terms of CMT, the conceptual metaphor underlying distance metaphors might be described as DURATION IS DISTANCE, for size metaphors it might be DURATION IS A SPECIFIC ENTITY’S PHYSICAL DIMENSIONS and for amount metaphors: DURATION IS A NUMBER OF DISTINCT ENTITIES.

For a language like English, which distinguishes between different quantifiers for mass and count nouns, it might be justifiable to discuss mass quantifiers under size metaphors and count quantifiers under number metaphors. Although time in Dutch is a mass noun (because in combination with a quantifier, the singular form is used), the plural form of tijd, tijden, is also frequent, but not in combination with quantifiers. Besides, Dutch does not distinguish count quantifiers from mass quantifiers: quantifiers have the same form, regardless whether the noun they are modifying is a mass noun or a count noun, compare (8a) and (8b):

(8) a. veel water much/many water ‘much water’ b. veel kind-eren much/many child-PL ‘many children’ c. veel tijd much/many time ‘much time’

For those reasons, I do not believe there is enough evidence to assume that Dutch duration metaphors including quantifiers, such as (8c)11, might only facilitate the link between mass amounts and duration. I am not trying to say that there is no difference in Dutch between count nouns and mass nouns. Conceptually there is a difference, and there are certain quantifiers that for that reason only go with count nouns, such as een paar ‘a few’. However, all quantifiers that can go with mass nouns can also go with count nouns. When trying to find implications for

(27)

cognitive representations based on linguistic structures, this is an important piece of data. That is to say, it implies that wherever a quantifier occurs in a Dutch expression with a mass noun, the representation of this quantifier in the brain includes quantifications of distinct entities as well. Given that tijd is such a mass noun occurring with quantifiers, it might very well be possible that the experience of time is also cognitively linked to amounts of distinct entities.

Although this threefold distinction between distance, size and amount metaphors will structure the discussion of Dutch duration metaphors, it is not the starting point for the investigation of these metaphors. The aim of the present section is to discuss to what extent such a distinction makes sense in view of the data. As discussed in section 3.1, the data for this investigation comes from SoNaR and CGN. However, the crucial point is how this data is investigated. If a corpus is searched for a specific expression, obviously, no expressions will be found that were not already known beforehand. If, in that way, a threefold distinction is the starting point for an investigation, only expressions that fit one of these three types will be found. To avoid such a bias towards a threefold distinction, firstly several n-gram investigations were carried out. For example, SoNaR was investigated for combinations of tijd preceded by any possible word. Outcomes were automatically ordered by decreasing frequencies. Subsequently, by qualitative examination of the outcomes, I decided which expressions in the frequency lists should be considered duration metaphors and which not. Sometimes, outcomes inspired new corpus investigations. For example, if both lange tijd ‘long time’ and langere tijd ‘longer time’ occur relatively frequently, it might be interesting to investigate the corpus for

lang (as lemma) tijd. In that way, the use of lange tijd as well as the use of langere tijd

(comparative) and langste tijd (superlative) is displayed. Moreover, also possible writing errors of the adjective are included in this way.

Unfortunately, only SoNaR has this n-gram search function. It is possible to search in the CGN for n-grams, but the outcomes are all listed as distinct hits, instead of frequencies grouped per n-gram. Therefore, the SoNaR n-gram investigations form the basis of this study, be it that comparisons are made with frequencies found in the CGN.

3.3. Analysis

Table 1 includes the frequencies of the 10 most frequent duration metaphors found in SoNaR by searching for ‘any word + tijd as lemma’, and their frequencies in the CGN.

(28)

SoNaR CGN Translation

lange tijd 12,531 114 long time

geen tijd 9770 297 no time

hele tijd 9212 599 all the time

korte tijd 7726 99 short time

enige tijd 6734 37 some time

veel tijd 5881 194 much time

meer tijd 5215 127 more time

geruime tijd 3891 20 quite some time

wat tijd 2527 58 some time

weinig tijd 2315 60 few time12

Table 1. Top 10 most frequent duration metaphors in SoNaR. Token frequencies in SoNaR and CGN

The expressions in table 1 are duration metaphors, since the words modifying tijd are also used in spatial sense when modifying nouns not related to the temporal domain. The only debatable expression in this respect is geruime tijd. Etymologically, geruime is derived from the spatial adjective ruim ‘wide’, and started being used in temporal sense, besides spatial sense, in

Vroegnieuwnederlands (Early Modern Dutch, 1500-1700). Since the 18th century its spatial use got lost (“ruim 2” 2003). According to the Dikke van Dale, nowadays, geruime is merely used as an attributive adjective modifying temporal nouns (“geruime” 2016). However, SoNaR includes at least nine instances of geruime afstand ‘considerable distance’, in clearly spatial sense, such as:

(9) Ze staan op geruime afstand van elkaar […]

PRO:3PL stand.PL on13 some distance from PRO:REC

‘They are located on considerable distance from each other […]’ (about the flowers of a particular bush)

Classifying the duration metaphors in table 1 as either distance, size or number metaphor is not completely straightforward. The expressions lange tijd and korte tijd are classified as distance metaphor, as they irrefutably include terminology that is also used to talk about space, namely the adjectives lang ‘long’ and kort ‘short’. Likewise, the following expressions can be classified as number metaphors right away: enige, veel, meer, wat and weinig tijd. Hele tijd could be classified as a size metaphor, although ‘proportional metaphor’ might perhaps be a more suitable descriptive term for this specific case. That leaves geen tijd, and geruime tijd as unclassified duration metaphors. As geen tijd describes a lack of duration, it could be argued that this is not a distance metaphor. However, if regarded as one, it would be classified as a

12_{The proper English translation of weinig tijd is ‘little time’. However, ‘little time’ arguably is a size metaphor,} but weinig tijd is not. To avoid confusion on that point, I have chosen the more literal translation with the unambiguous quantifier ‘few’, even though this quantifier does not modify mass nouns in English.

13_{Dutch makes widespread use of prepositions. I prefer to gloss prepositions with a lexical gloss instead of a} common grammatical gloss PREP, to highlight this wealth of prepositions. Although in different contexts, a Dutch preposition might be translated best with different English prepositions, for the sake of consistency (and to do justice to the Dutch prepositional system), every preposition is glossed with the same lexical gloss throughout this thesis. The English translation that is added to each Dutch example contains the preferred English translation of the Dutch preposition for the specific context. For an extensive discussion of the different senses of several high frequent Dutch prepositions, I refer to Colombo and Floris d’Arcais (1984).

(29)

number metaphor. Since the status of geruime tijd as a duration metaphor is debatable, I will leave it unclassified here. On the basis of this classification, the relative frequencies of the duration metaphors in the different groups show that number metaphors are more frequent than distance metaphors in both SoNaR and the CGN, even if geen tijd is not included as number metaphor (see figure 1).

Figure 1 shows the relative frequencies in SoNaR and CGN of the different groups of duration metaphors discussed above. These only include the frequencies of the top 10 most frequent duration metaphors including the word tijd, as described in table 1. Based on these data, it is impossible to conclude which type of duration metaphor is most prevalent in Dutch. While in SoNaR, distance and number metaphors are almost equally frequent (respectively 30.8% and 34.5%), in CGN, number metaphors are clearly more frequent than distance metaphors (29.7% compared to 13.3%). Then there is the issue of size metaphors, which are, if hele tijd is regarded as one, the most prevalent metaphors in CGN. Altogether, merely looking at the most frequent duration metaphors including the word tijd, does not provide sufficient evidence to describe the preferences of duration metaphors in Dutch. Therefore, the three different types of duration metaphors are discussed in more detail below, including moreover, investigations of durational metaphors that do not contain the word tijd.

3.3.1. Distance metaphors

The only distance metaphors mentioned in the previous section were lange tijd and korte tijd, yet another prevalent Dutch distance metaphor is found in the use of lang as a temporal adverb, for example as in (10):

(10) Het kan nog lang dur-en voor het genezen is.

PRO:3SG.N AUX.PRS.3SG yet long last-INF for PRO:3SG.N heal.PP AUX.PRS.3SG

‘It may still take long before it is healed.’ (SoNaR) 30.8 34.5 14.8 14.0 5.9 SoNaR Distance metaphors Number metaphors geen tijd hele tijd geruime tijd 13.3 29.7 18.5 37.3 1.2 CGN

Figure 1. Distribution of SoNaR’s top 10 most frequent duration metaphors in SoNaR and CGN in percentage of total tokens.

(30)

Investigations of n-grams for ‘lemma: lang + lemma: duren’ and for ‘lemma: duren + lemma:

lang’, reveal that this duration metaphor occurs 5672 times in SoNaR and 201 times in CGN.

Its counterparts with kort occur less frequent in both corpora, with 164 hits in SoNaR and 16 hits in the CGN. However, many instances of lang duren are actually instances of its negation:

niet lang duren ‘not take long’. Actually, this negation accounts for 534 of the 5672 SoNaR

hits, and 37 of the CGN hits. Still, when this is taken into account, the ‘long distance metaphor’ occurs more often than the ‘short distance metaphor’.

Widely used is the adverb lang modifying geleden, the past participle of the verb lijden ‘to pass’: lang geleden. This verb does not occur in temporal sense any more in present day Dutch, except for in frozen construction, such as lang geleden ‘long ago’, see (11):

(11) Lang geleden heeft hij een ander boek gepubliceerd,

long ago AUX.PRS.3SG PRO:3SG.M INDEF other book publish.PP toen hij nog naar de universiteit ging.

when PRO:3SG.M still to DEF.C university go.PST.3SG

‘Long ago, he published another book, when he was still in college.’ (SoNaR)

Uses of lang geleden referring to a remote past as in (11) are most frequent in SoNaR. However, we need to be careful with taking the frequency count of lang geleden as an indicator of the frequency of lang geleden as a distance metaphor for a long durational distance in the past. As it happens, of the 9294 hits of lang geleden, at least 1772 are actually negations of this expression, indicating a short duration. See, for example:

(12) Het project is nog niet zo lang geleden stopgezet.

DEF.N project be.PRS.3SG yet NEG so long ago stop.PP

‘The project was stopped not yet that long ago.’ (SoNaR)

Actually, a negation of lang geleden occurs more frequently to indicate a short duration than its antonym kort geleden ‘short ago’.14_{To be precise, SoNaR includes merely 791 instances}

of kort geleden, of which only 1 is a negation of kort geleden, presented in example (13). In fact, this one instance of niet zo kort geleden actually seems to be a mistake, a permutation of the words niet en nog or an accidental use of kort instead of lang:

14_{As a native speaker of Dutch, intuitively I would say that niet (zo) lang geleden usually indicates a remoter past} than kort geleden in absolute temporal sense. However, niet (zo) lang geleden will be used when the subjective experience of the time span is perceived to be short. In that sense, niet (zo) lang geleden and kort geleden are not semantically, or pragmatically, interchangeable. The relative low frequency of kort geleden compared to lang

geleden, should thus not be explained by stating that kort geleden and niet (zo) lang geleden should be grouped

together as antonym of lang geleden. Rather, the difference in frequency might be explained by the presence of several temporal adverbs in Dutch to describe a short timespan into the past, such as recentelijk, onlangs and pas

(geleden) ‘recently’. Whereas, for lang geleden, merely one temporal adverb could be used as alternative: vroeger

Do longer lines take longer? Reconsidering the cognitive reflections of spatial duration metaphors: Evidence from Dutch